• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권3호
• /
• pp.837-852
• /
• 2021

Recurrent neural network (RNN) architectures have been used for language modeling (LM) tasks that require learning long-range word or character sequences. However, the RNN architecture is still suffered from unstable gradients on long-range sequences. To address the issue of long-range sequences, an attention mechanism has been used, showing state-of-the-art (SOTA) performance in all LM tasks. A differentiable neural computer (DNC) is a deep learning architecture using an attention mechanism. The DNC architecture is a neural network augmented with a content-addressable external memory. However, in the write operation, some information unrelated to the input word remains in memory. Moreover, DNCs have been found to perform poorly with low numbers of weight parameters. Therefore, we propose a robust memory deallocation method using a limited retention vector. The limited retention vector determines whether the network increases or decreases its usage of information in external memory according to a threshold. We experimentally evaluate the robustness of a DNC implementing the proposed approach according to the size of the controller and external memory on the enwik8 LM task. When we decreased the number of weight parameters by 32.47%, the proposed DNC showed a low bits-per-character (BPC) degradation of 4.30%, demonstrating the effectiveness of our approach in language modeling tasks.

• Animal cells and systems
• /
• 제13권3호
• /
• pp.331-337
• /
• 2009

Twenty-eight Anopheles species has been so-far identified in Iran, while only 8 species was proved as malaria vector. In this study, we principally examined the cuticular hydrocarbon (CHC) potency in identification of Iranian vectors of malaria and then differentiation of vector and non-vector species of Anopheles. Seven species of malaria vectors and the non-vector species, Anopheles claviger were collected throughout Iran. Female extracts were made out of every five conspecific specimens by surface immersion in pure n-hexane. Each sample was injected into a FID-GC instrument along with the known concentrations of standards. CHC profiles of the eight Anopheles species indicated no qualitative difference. The average mass of each eluted CHC were compared using Repeated ANOVA and Mann-Whitney tests. Results confirmed a significant difference in mass of each single CHC at a specific retention time (RT). Statistical comparison of CHC mass in An. sacharovi, An. stephensi, An. culicifacies and An. fluviatilis at RT 39.6 indicated significant differences (P<0.05) among these species. Analysis of CHC mass of An. dthali, An. superpictus & An. sacharovi at RT 28.5, An. stephensi & An. sacharovi at RT 30.7 and An. sacharovi & An. claviger at RT 30.6 similarly indicated significant differences (P<0.05). An. sacharovi could be distinguished from other species, which showed only trace, by integratable peaks at retention times of 29.7, 31 and 32.6. Similarly, An. claviger could be distinguished from the other species with a trace peak at RT 30.6. In order to separate An. stephensi from the five other species, the integratable peak at RT 30.7 was used. An. dthali could be identified at RT 26.2 by an integratable peak v.s. the trace peaks of other species. An. superpictus had indicator peaks at RTs 27.4 & 28.5 v.s. trace peaks of other species. An. maculipennis with its trace peak at RT 39.6 could be easily differentiated from An. fluviatilis & An. culicifacies. This study proved that all of the examined species of Anopheles could be well identified based on their quantitative differences in CHCs, except for An. fluviatilis & An. culicifacies for which no CHC indicator peak was detected.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제14권4호
• /
• pp.1562-1578
• /
• 2020

Vector data compression algorithm can meet requirements of different levels and scales by reducing the data amount of vector graphics, so as to reduce the transmission, processing time and storage overhead of data. In view of the fact that large threshold leading to comparatively large error in Douglas-Peucker vector data compression algorithm, which has difficulty in maintaining the uncertainty of shape features and threshold selection, a segmented Douglas-Peucker algorithm based on node importance is proposed. Firstly, the algorithm uses the vertical chord ratio as the main feature to detect and extract the critical points with large contribution to the shape of the curve, so as to ensure its basic shape. Then, combined with the radial distance constraint, it selects the maximum point as the critical point, and introduces the threshold related to the scale to merge and adjust the critical points, so as to realize local feature extraction between two critical points to meet the requirements in accuracy. Finally, through a large number of different vector data sets, the improved algorithm is analyzed and evaluated from qualitative and quantitative aspects. Experimental results indicate that the improved vector data compression algorithm is better than Douglas-Peucker algorithm in shape retention, compression error, results simplification and time efficiency.

• 한국환경과학회지
• /
• 제16권10호
• /
• pp.1169-1177
• /
• 2007

The main objective of this research was to evaluate the effects of process variables which were forming ability, flow displacement, effective stress, effective strain, fluid vector and products defects on manufactured artificial lightweight aggregate made of both incinerated sewage sludge ash and clay by means of the numerical analysis of a rigid-plastic finite element method. CATIA (3D CAD program) was used for an extrusion metal mold design that was widely used in designing aircraft, automobile and metallic molds. A metal forming analysis program (ATES Co.) had a function of a rigid-plastic finite element method was used to analyze the program. The result of extrusion forming analysis indicated clearly that a shape retention of the manufactured artificial light-weight aggregate could be maintained by increasing the extrusion ratio (increasing compressive strength inside of extrusion die) and decreasing the die angle. The stress concentration of metal mold was increased by increasing an extrusion ratio, and it was higher in a junction of punch and materials, friction parts between a bottom of the punch and inside of a container, a place of die angle and a place of die of metal mold. Therefore, a heat treatment as well as a rounding treatment for stress distribution in the higher stress concentration regions were necessary to extend a lifetime of the metallic mold. A deformity of the products could have made from several factors which were a surface crack, a lack of the shape retention and a crack of inside of the products. Specially, the surface crack in the products was the most notably affected by the extrusion ratio.

• 지능정보연구
• /
• 제19권2호
• /
• pp.125-140
• /
• 2013

어떤 클래스에 속한 레코드의 개수가 다른 클래스들에 속한 레코드의 개수보다 매우 많은 경우에, 이 데이터 집합을 '불균형 데이터 집합'이라고 한다. 데이터 분류에 사용되는 많은 기법들은 이러한 불균형 데이터에 대해서 저조한 성능을 보인다. 어떤 기법의 성능을 평가할 때에 적중률뿐만 아니라, 민감도와 특이도도 함께 측정하여야 한다. 고객의 이탈을 예측하는 문제에서 '유지' 레코드가 다수 클래스를 차지하고, '이탈' 레코드는 소수 클래스를 차지한다. 민감도는 실제로 '유지'인 레코드를 '유지'로 예측하는 비율이고, 특이도는 실제로 '이탈'인 레코드를 '이탈'로 예측하는 비율이다. 많은 데이터 마이닝 기법들이 불균형 데이터에 대해서 저조한 성능을 보이는 것은 바로 소수 클래스의 적중률인 특이도가 낮기 때문이다. 불균형 데이터 집합에 대처하는 과거 연구 중에는 소수 클래스를 Oversampling하여 균형 데이터 집합을 생성한 후에 데이터 마이닝 기법을 적용한 연구들이 있다. 이렇게 균형 데이터 집합을 생성하여 예측을 수행하면, 특이도는 다소 향상시킬 수 있으나 그 대신 민감도가 하락하게 된다. 본 연구에서는 민감도는 유지하면서 특이도를 향상시키는 모델을 개발하였다. 개발된 모델은 Support Vector Machine (SVM), 인공신경망(ANN) 그리고 의사결정나무 기법 등으로 구성된 하이브리드 모델로서, Hybrid SVM Model이라고 명명하였다. 구축과정 및 예측과정은 다음과 같다. 원래의 불균형 데이터 집합으로 SVM_I Model과 ANN_I Model을 구축한다. 불균형 데이터 집합으로부터 Oversampling을 하여 균형 데이터 집합을 생성하고, 이것으로 SVM_B Model을 구축한다. SVM_I Model은 민감도에서 우수하고, SVM_B Model은 특이도에서 우수하다. 입력 레코드에 대해서 SVM_I와 SVM_B가 동일한 예측치를 도출하면 그것을 최종 해로 결정한다. SVM_I와 SVM_B가 상이한 예측치를 도출한 레코드에 대해서는 ANN과 의사결정나무의 도움으로 판별 과정을 거쳐서 최종 해를 결정한다. 상이한 예측치를 도출한 레코드에 대해서는, ANN_I의 출력값을 입력속성으로, 실제 이탈 여부를 목표 속성으로 설정하여 의사결정나무 모델을 구축한다. 그 결과 다음과 같은 2개의 판별규칙을 얻었다. 'IF ANN_I output value < 0.285, THEN Final Solution = Retention' 그리고 'IF ANN_I output value ${\geq}0.285$, THEN Final Solution = Churn'이다. 제시되어 있는 규칙의 Threshold 값인 0.285는 본 연구에서 사용한 데이터에 최적화되어 도출된 값이다. 본 연구에서 제시하는 것은 Hybrid SVM Model의 구조이지 특정한 Threshold 값이 아니기 때문에 이 Threshold 값은 대상 데이터에 따라서 얼마든지 변할 수 있다. Hybrid SVM Model의 성능을 UCI Machine Learning Repository에서 제공하는 Churn 데이터 집합을 사용하여 평가하였다. Hybrid SVM Model의 적중률은 91.08%로서 SVM_I Model이나 SVM_B Model의 적중률보다 높았다. Hybrid SVM Model의 민감도는 95.02%이었고, 특이도는 69.24%이었다. SVM_I Model의 민감도는 94.65%이었고, SVM_B Model의 특이도는 67.00%이었다. 그러므로 본 연구에서 개발한 Hybrid SVM Model이 SVM_I Model의 민감도 수준은 유지하면서 SVM_B Model의 특이도보다는 향상된 성능을 보였다.

• BMB Reports
• /
• 제30권1호
• /
• pp.60-65
• /
• 1997

Expression of human Cu.Zn-superoxide dismutase (SOD) with activity comparable to human erythrocyte enzyme was achieved in E. coli B21(DE3) by using the pET-17b expression vector containing a T7 promoter. Recombinant human SOD was found in the cytosol of disrupted bacterial cells and represented > 25% of the total bacterial proteins. The protein produced by the E. coli cells was purified using a combination of ammonium sulfate precipitation, Sephacryl S-100 gel filtration and DEAE-Sephacel ion exchange chromatography. The recombinant Cu,Zn-SOD and human erythrocyte enzyme were compared using dismutation activity, SDS-PAGE and immunoblotting analysis. The mass of the subunits was determined to be 15,809 by using a electrospray mass spectrometer. The copper specific chelator. diethyldithiocarbamate (DOC) reacted with the recombinant Cu,Zn-SOD. At $50{\mu}M$ and $100{\mu}M$ concentrations of DOC, the dismutation activity was not inhibited for one hour but gradually reduced after one hour. This result suggests that the reaction of DOC with the enzyme occurred in two distinct phases (phase I and phase II). During phase I of this reaction, one DOC reacted with the copper center, with retention of the dismutation activity while the second DOC displaced the copper, with a loss of activity in phase II.

• Parasites, Hosts and Diseases
• /
• 제49권4호
• /
• pp.357-364
• /
• 2011

Various Leishmania species were engineered with green fluorescent protein (GFP) using episomal vectors that encoded an antibiotic resistance gene, such as aminoglycoside geneticin sulphate (G418). Most reports of GFP-Leishmania have used the flagellated extracellular promastigote, the stage of parasite detected in the midgut of the sandfly vector; fewer studies have been performed with amastigotes, the stage of parasite detected in mammals. In this study, comparisons were made regarding the efficiency for in vitro G418 selection of GFP-Leishmania amazonensis promastigotes and amastigotes and the use of in vivo G418 selection. The GFP-promastigotes retained episomal plasmid for a prolonged period and G418 treatment was necessary and efficient for in vitro selection. In contrast, GFP-amastigotes showed low retention of the episomal plasmid in the absence of G418 selection and low sensitivity to antibiotics in vitro. The use of protocols for G418 selection using infected BALB/c mice also indicated low sensitivity to antibiotics against amastigotes in cutaneous lesions.

• Journal of Microbiology and Biotechnology
• /
• 제18권5호
• /
• pp.866-873
• /
• 2008

2-Deoxystreptamine is a core aglycon that is vital to backbone formation in various aminoglycosides. This core structure can be modified to develop hybrid types of aminoglycoside antibiotics. We obtained three genes responsible for 2-deoxystreptamine production, neo7, neo6, and neo5, which encode 2-deoxy-scyllo-inosose synthase, L-glutamine: 2-deoxy-scyllo-inosose aminotransferase, and dehydrogenase, respectively, from the neomycin gene cluster. These genes were cloned into pIBR25, a Streptomyces expression vector, resulting in pNDOS. The recombinant pNDOS was transformed into a non-aminoglycoside-producing host, Streptomyces venezuelae YJ003, for heterologous expression. Based on comparisons of the retention time on LC-ESI/MS and ESI-MS data with those of the 2-deoxystreptamine standard, a compound produced by S. venezuelae YJ003/pNDOS was found to be 2-deoxystreptamine.

• Bulletin of the Korean Chemical Society
• /
• 제22권1호
• /
• pp.46-52
• /
• 2001

To evaluate the non-ionic polymer, poly(ethylene glycol) (PEG), as a component in cationic copolymers for non-viral gene delivery systems, PEG was coupled to polyethylenimine (PEI). We present the effects of different degrees and shapes of pegylation of PEI on cytotoxicity, water solubility and transfection efficiency. This work reports the synthesis and characterization of a series of cationic copolymers on the basis of the conjugates of PEI with PEG. The modified molecules were significantly less toxic than the original polymer. Moreover, the chemical modification led to enhancement of their solubility. The comparison of pegylated PEIs with different degrees of derivation showed that all the polymers tested reached comparable levels of transgene expression to that of native PEI. As assessed by agarose gel electrophoresis, even highly substituted PEI derivatives were still able to form polyionic complexes with DNA. However, aside from an increase in solubility and retention of the ability to condense DNA, methoxy-PEG-modified PEIs resulted in a significant decrease in the transfection activity of the DNA complexes. In fact, the efficiency of the copolymer was compromised even at a low degree of modification suggesting that the PEG action resulting from its shape is important for efficient gene transfer. The mode of PEG grafting and the degree of modification influenced the transfection efficiency of PEI.

• Molecules and Cells
• /
• 제44권10호
• /
• pp.770-779
• /
• 2021

Transgenic Arabidopsis thaliana expressing an anti-rabies monoclonal antibody (mAb), SO57, was obtained using Agrobacterium-mediated floral dip transformation. The endoplasmic reticulum (ER) retention signal Lys-Asp-Glu-Leu (KDEL) was tagged to the C-terminus of the anti-rabies mAb heavy chain to localize the mAb to the ER and enhance its accumulation. When the inaccurately folded proteins accumulated in the ER exceed its storage capacity, it results in stress that can affect plant development and growth. We generated T₁ transformants and obtained homozygous T₃ seeds from transgenic Arabidopsis to investigate the effect of KDEL on plant growth. The germination rate did not significantly differ between plants expressing mAb SO57 without KDEL (SO plant) and mAb SO57 with KDEL (SOK plant). The primary roots of SOK agar media grown plants were slightly shorter than those of SO plants. Transcriptomic analysis showed that expression of all 11 ER stress-related genes were not significantly changed in SOK plants relative to SO plants. SOK plants showed approximately three-fold higher mAb expression levels than those of SO plants. Consequently, the purified mAb amount per unit of SOK plant biomass was approximately three times higher than that of SO plants. A neutralization assay revealed that both plants exhibited efficient rapid fluorescent focus inhibition test values against the rabies virus relative to commercially available human rabies immunoglobulins. KDEL did not upregulate ER stress-related genes; therefore, the enhanced production of the mAb did not affect plant growth. Thus, KDEL fusion is recommended for enhancing mAb production in plant systems.